The efficacy of therapeutic granulocyte transfusions is limited by the relatively small number of cells obtained using standard starch and steroid stimulation of the donor. We evaluated the safety and efficacy of daily granulocyte colony stimulating factor (G-CSF)-stimulated granulocyte transfusions in two patients, a 55 kg female with T cell large granular lymphocytic leukemia and a sigmoid phlegmon, and a 131 kg male with systematic aspergillus flavus infection on day 10 following T cell depleted marrow allo-grafting for myeloma. Both patients had ANCs <0.1x109/L, and were receiving G-CSF 10 microgram/kg/d and systemic antibiotics and antifungal agents. Healthy volunteer allogeneic apheresis donors were given a single dose of G-CSF 5 microgram/kg SQ 12 to 24 hours prior to apheresis as well as a single dose of dexamethasone 8 mg orally 12 hours prior to donation. Seven liters of whole blood were processed on the Fenwal CS-3000 device using Hetastarch as the sedimenting agent. A different non-HLA matched donor was used each day. Four irradiated granulocyte transfusions were given to the first patient and 12 to the second patient prior to recovery of the patients' own neutrophils. Mean donor white blood cell count prior to apheresis (m + sd) was 29.9 + 3.8 x109/L (94% granulocytes), range 25.1-37.6 x109/L. The granulocytapheresis products contained 7.4 + 1.3 x 1010 granulocytes in a mean volume of 379 mL. Platelet content of the granulocyte products was 5.0 x 1011, equivalent to a 9-unit platelet transfusion. The mean number of granulocytes transfused per kg was 1.2 x 109/kg in the first and 0.6 x 109 /kg in the second patient. Dramatic one-hour post-transfusion increments in circulating neutrophil counts were routinely seen in both patients (immediate ANC increments of 1.99 and 1.87 x109/L, respectively). ANCs continued to rise for up to 8 hours after each transfusion and did not decline to pre-transfusion levels until 24-30 hours after infusion. Both patients became afebrile and exhibited stabilization and/or improvement in clinical, hemodynamic, and radiographic indicators of infection. The granulocyte transfusions were well tolerated, and co-administration (within 12 hours) of amphotericin did not lead to pulmonary compromise. Neither patient became HLA-alloimmunized. Donors experienced mild bone pain, headache, and insomnia but were willing and eager to support such donations again. In summary, G-CSF-recruited granulocyte components contained four-fold greater numbers of cells than historical non-G-CSF mobilized products, were well tolerated, and appeared to have a markedly prolonged circulatory half-life, perhaps due to G-CSF-related inhibition of apoptosis or recruitment of younger granulocytes. The beneficial effects experienced in our two patients suggest that this new transfusion component deserves broader study.